Research on Face Image Digital Processing and Recognition Based on Data Dimensionality Reduction Algorithm

Effect of Ambient Lights on the Accuracy of a 3-Dimensional Optical Scanner for Face Scans: An In Vitro Study

Most 3D scanners use optical technology that is impacted by lighting conditions, especially in triangulation with structured-light or laser techniques. However, the effect of ambient lights on the accuracy of the face scans remains unclear. The purpose of this study is to investigate the effect of ambient lights on the accuracy of the face scans obtained from the face scanner (EinScan Pro 2X Plus, Shining 3D Tech. Co., LTD., Hangzhou, China). A head model was designed in Rhinoceros 5 software (Rhino, Robert McNeel and Associates for Windows, Washington DC, USA) and printed with 200 micron resolution of polylactic acid and was dented with 2.0 mm of carbide bur to aid in superimposition in software. The head model was measured by a coordinate-measuring machine (CMM) to generate a reference stereolithography (STL) file as a control. The face model was scanned four times under nine light conditions: cool white (CW), warm white (WW), daylight (DL), natural light (NL), and illuminant (9w, 18w, 22w). Scan data were exported into an STL file. The scan STL files obtained were compared with the reference STL file by 3D inspection software (Geomagic Control X version 17, Geomagic, Morrisville, NC, USA). The deviations and root mean square errors (RMSEs) between the reference model (trueness) and within the group (precision) were selected for the statistical analysis. The statistical analysis was done using SPSS 20.0 (IBM Company, Chicago, USA). The trueness and precision were evaluated with the one-way ANOVA with multiple comparisons using the Tukey method. For trueness, the scanner showed the lowest RMSE under the NL group (77.18 ± 3.22) and the highest RMSE under the 18w-DL group (95.33 ± 6.89). There was a statistically significant difference between the NL group and the 18w-DL group (p < 0.05) for trueness. Similarly, for precision, the scanner showed the lowest RMSE under the NL group (56.92 ± 4.56) and the highest RMSE under the 9w-CW group (78.52 ± 10.61). There was statistically significant difference between NL, 18w-WW, 18w-CW, 18w-DL, 22w-WW, 22w-DL, 9w-CW, 9w-WW, and 9w-DL (p < 0.05) for the precision. Ambient lights affected the face scans. Under the natural light condition, the face scanner had the best accuracy in terms of both trueness and precision. The 18w-DL and 9w-WW conditions showed the least trueness whereasthe 9w-CW and 9w-DL conditions showed the least precision.

Deep Neural Networks for Automatic Flower Species Localization and Recognition

Deep neural networks are efficient methods of recognizing image patterns and have been largely implemented in computer vision applications. Object detection has many applications in computer vision, including face and vehicle detection, video surveillance, and plant leaf detection. An automatic flower identification system over categories is still challenging due to similarities among classes and intraclass variation, so the deep learning model requires more precisely labeled and high-quality data. In this proposed work, an optimized and generalized deep convolutional neural network using Faster-Recurrent Convolutional Neural Network (Faster-RCNN) and Single Short Detector (SSD) is used for detecting, localizing, and classifying flower objects. We prepared 2000 images for various pretrained models, including ResNet 50, ResNet 101, and Inception V2, as well as Mobile Net V2. In this study, 70% of the images were used for training, 25% for validation, and 5% for testing. The experiment demonstrates that the proposed Faster-RCNN model using the transfer learning approach gives an optimum mAP score of 83.3% with 300 and 91.3% with 100 proposals on ten flower classes. In addition, the proposed model could identify, locate, and classify flowers and provide essential details that include flower name, class classification, and multilabeling techniques.